In snfagora/autotextclassifier: R Package for Automatically Classifying Texts Based on Tidymodels

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  rmarkdown.html_vignette.check_title = FALSE
)

Load libraries

library(autotextclassifier) # Auto text classifier 
library(parallel) # Parallel processing 
library(doParallel) # Parallel processing 
library(here) # Creating reproducible file paths 
library(patchwork) # Putting ggplots together
library(recipes) # Preprocessing 
library(zeallot) # Multiple assignments 
library(yardstick) # Metrics  

Import data

load(file = here("inst/extdata/sample_data.rda"))

Data munging

Don't forget to make sure that the type of the outcome variable should be factor.

names(sample_data) <- c("category", "org_name", "ein", "text")
sample_data$category <- as.factor(sample_data$category)

Apply basic recipe

The rec object provides the following information. The function also checks whether the text column has missing values or includes extremely short documents (less than five words).

There are two broad basic options for text preprocessing.

1. Without word embedding

2. Tokenization for text [trained]

3. Stop word removal for text [trained]
4. Text filtering for text [trained]

5. Term frequency-inverse document frequency with text [trained]

6. With word embedding

7. Tokenization for text [trained]

8. Stop word removal for text [trained]
9. Text filtering for text [trained]
10. Word embeddings aggregated from text [trained]

# Without word embedding 
rec <- apply_basic_recipe(sample_data, category ~ text, text)

# With word embedding 
rec_alt <- apply_basic_recipe(sample_data, category ~ text, text, add_embedding = TRUE)

Split data

set.seed(1234)

c(train_x_class, test_x_class, train_y_class, test_y_class) %<-% split_using_srs(input_data = sample_data, category = category, rec = rec)

Create tuning parameters

c(lasso_spec, rand_spec, xg_spec) %<-% create_tunes()

Create search spaces

c(lasso_grid, rand_grid, xg_grid) %<-% create_search_spaces(train_x_class, category, lasso_spec, rand_spec, xg_spec)

Create workflows

c(lasso_wf, rand_wf, xg_wf) %<-% create_workflows(lasso_spec, rand_spec, xg_spec, category)

Create 10-fold cross-validation samples

set.seed(1234)

class_folds <- create_cv_folds(train_x_class, train_y_class, category)

Find the best model from each algorithm

Consider using parallel processing to speed up.

all_cores <- parallel::detectCores(logical = FALSE)

cl <- makeCluster(all_cores[1] - 1)

registerDoParallel(cl)

The default metric for model evaluation is accuracy. The other options are balanced accuracy (bal_accuracy), F-score (f_means), and Area under the ROC curve (roc_auc).

c(best_lasso, best_rand, best_xg) %<-% find_best_model(lasso_wf, rand_wf, xg_wf, class_folds, lasso_grid, rand_grid, xg_grid, metric_choice = "accuracy")

Fit the best model from each algorithm to the data

This step would take the longest running time. We recommend saving this output as an RData object somewhere.

c(lasso_fit, rand_fit, xg_fit) %<-% fit_best_model(lasso_wf, best_lasso, rand_wf, best_rand, xg_wf, best_xg, train_x_class, train_y_class, category)

Evaluate the model using visualization

# Based on the class-based metrics 

viz_class_fit(lasso_fit, "Lasso", test_x_class, test_y_class, "class") + 

viz_class_fit(rand_fit, "Random forest", test_x_class, test_y_class, "class") +

viz_class_fit(xg_fit, "XGBoost", test_x_class, test_y_class, "class")

# Based on the probability-based metrics 

viz_class_fit(lasso_fit, "Lasso", test_x_class, test_y_class, "probability") +

viz_class_fit(rand_fit, "Random forest", test_x_class, test_y_class, "probability") +

viz_class_fit(xg_fit, "XGBoost", test_x_class, test_y_class, "probability")

Predict new data

This is just a hypothetical example. Let's denote the new data as "new_sample_data." (It has the data structure as the sample_data.) The tidymodels package provides the following functions.

# Preprocessding data using the recipe created earlier 
new_preporcessed <- bake(rec, new_sample_data)

# Making predictions using the best lasso model 
new_sample_data$category <- predict(lasso_fit, new_preporcessed)$.pred_class 

# If you want to extract probabilities, use the following depending on whether you want to predict FALSE or TRUE values.
# new_sample_data$category <- predict(lasso_fit, new_preporcessed, type = "prob")$.pred_FALSE.
# new_sample_data$category <- predict(lasso_fit, new_preporcessed, type = "prob")$.pred_TRUE.

snfagora/autotextclassifier documentation built on June 16, 2022, 9:42 a.m.